Dear Aventine Readers, 

We’ve written a lot about advances in well-known clean energy sources like solar, wind and nuclear in this newsletter. This month we look at one that has gotten far less attention: next-generation geothermal power. While traditional geothermal energy taps naturally occurring underground wells of hot water to drive turbines and create electricity, the power it can produce is limited by the finite number of such wells. Next-generation geothermal doesn’t face such limitations because all it needs is hot rock, which is reliably found all over the planet far below the Earth’s surface. There are significant challenges, including the need for entirely new technologies to tap into these deep energy reserves. But if early demonstrations are successful, this next wave of geothermal energy could be transformative.

Also in this issue: How the urgency to create a vaccine for HIV is colliding with life-saving breakthroughs in the improvement of pre-exposure prophylaxis (PrEP) drugs; how AI is being used in hospital operating rooms to document and assess the many decisions and actions taken during medical procedures; and a push to decarbonize ocean freight by harnessing wind power. 

Thanks for reading and I hope you are all enjoying your summer.

Danielle Mattoon
Executive Director, Aventine

Subscribe

Subscribe to our newsletter and be kept up to date on upcoming Aventine projects

Dig a hole a mile deep and the bottom will likely be about 80°F hotter than the top. Keep going, and you’ll find temperatures that could solve the climate crisis. 

Modern geothermal power has been a small but constant source of clean energy throughout much of the world for the last 50 years or so, using naturally occurring wells of hot water beneath the Earth’s surface to drive turbines and create electricity. But such geological conditions are rare, which is why geothermal energy only contributes about 3.7 gigawatts of power to the U.S. grid today, or about 0.4 percent of America’s total electricity supply — and 16 gigawatts for the entire globe. 

But now, thanks to techniques borrowed from the oil and gas industry and recently developed technologies, a new, potentially transformative “next generation” of geothermal plants is taking shape. These systems drill into hot, dry rock at temperatures from 175°C (about 350°F) and potentially above 374°C (about 700°F), pump water down and use the hot water that comes back up to create power. Because subsurface hot rock exists everywhere, the technology could dramatically increase the geothermal resources available globally, and the United States is particularly well positioned to develop it. “The U.S. just happens to be a world leader in conventional geothermal and a world leader in oil and gas skill sets that are all directly aligned with advancing this sector,” said Lauren Boyd, director of the geothermal technologies office at the Department of Energy. “So to me it's an absolute no-brainer.”

Early estimates show that in the U.S., next-generation geothermal could unlock more than 5.5 terawatts of power, or more than five times the current U.S. power generation capacity. And that power could start coming online soon. Modeling by the National Renewable Energy Laboratory and Princeton University predicts that 30 gigawatts of next-generation geothermal could be added to the U.S. grid by 2030, with that figure rising to 100 gigawatts by 2050. Additionally, geothermal power can be generated continuously — and even shut off when not needed — and doesn’t require huge swaths of land as wind and solar do.

Not that it will be easy. While there is progress drilling into temperatures in the lower range of what’s needed for next-generation geothermal, tools and electronics don’t yet exist to survive the kind of heat that will deliver the highest levels of power and efficiency. Unforeseen geological conditions could derail entire projects. And the first of the new geothermal plants are enormously expensive to get up and running, with start-ups seeking initial funding in the range of several hundred million dollars. Despite the technology’s promise and need for capital, President Biden’s Bipartisan Infrastructure Law committed just $74 million to next-gen geothermal pilot projects, leaving much of the burden on private investment. 

Aventine spoke with startup leaders, academics and government officials about what’s ahead for the technology. They described its huge potential to provide vast quantities of energy along with a series of technical challenges that range in complexity. They also warned of the need for a high tolerance for risk among financial backers — and of the huge pressure on prototype plants to perform.

Getting steam out of stone

There are competing approaches for extracting heat from rock but they are all grounded in the basic principle that the hotter the rock, the greater the power and efficiency. And while geological idiosyncrasies give rise to areas of relatively shallow hot rock — and these would certainly be tapped first — much of geothermal’s potential energy is deeper below the surface, where the rock gets predictably hotter the deeper you go. “[Being able to drill] between five and 10 kilometers starts to get 50 percent of the world population with geothermal,” said Carlos Araque, co-founder and CEO of Quaise Energy. “From 10 to 20 kilometers gets upwards of 90 percent of the world's population with geothermal.” 

Of the various approaches being explored, two main ones stand out. The first, known as enhanced geothermal, uses well-established hydraulic fracturing — or fracking — techniques taken from the oil and gas playbook. Two long wells are drilled into hot rock, then high-pressure water containing sand is pumped down to create cracks that are propped open by the sand grains. This gives water a route through the hot and otherwise impermeable rock from one well to the other. Next, cold water is pumped down one well, rising in temperature as it travels through the rock and then returns up the second well to the surface, where it is used to drive a steam turbine to create electricity. This approach is relatively well understood, though the frequently documented seismic activity associated with oil and gas fracking looms large over it as a technology. Though the conditions created by fracking for geothermal are considered less problematic than those created for oil and gas extraction, the DOE has a protocol for seismic activity that all next-gen geothermal facilities in the U.S. must adhere to, said Boyd. 

The second approach, known as advanced geothermal, typically uses a similar two-well arrangement, but connects them via a series of narrow tunnels drilled through the rocks that pass water between the hot and cold wells. As the water travels through this network of tunnels, it's heated by the rock. “We’ve got to go down four or five kilometers, across four or five kilometers and hit an eight inch target in solid rock,” said John Redfern, CEO of the advanced geothermal startup Eavor. “That's difficult.” The company uses a technology called magnetic ranging, often used to help oil and gas drilling operations avoid one another, to ensure that its drilling operations meet. The company claims its approach, which doesn’t create networks of cracks and involves sealing the drilled holes so water is never in direct contact with the rock, also reduces the likelihood of seismic activity. 

While neither approach is ready to produce energy at scale, both are progressing quickly. Researchers at Utah FORGE, an underground laboratory backed by the Department of Energy for developing enhanced geothermal systems, have sent water between two wells via fractures using the enhanced geothermal technique. A startup called Fervo Energy (which did not respond to requests for an interview) completed a trial using the same fracking technique that successfully generated electricity, and is now drilling wells for a 400-megawatt plant next to the Utah FORGE site, with the first 70 megawatts of capacity expected to be online in 2026. In late June, Southern California Edison announced that it would purchase 320 megawatts of power from this plant as part of a 15-year agreement.

Meanwhile, Eavor has used the advanced geothermal technique to test a small closed-loop system in Canada and is currently drilling wells for its first plant, which the company says will generate 8 megawatts of electricity in Germany by 2027. There are other systems in development too. Notably, Sage Geosystems is developing systems in Texas that would use renewable energy to pump water down enhanced geothermal wells during times of high wind or solar production, and then release the heated water to generate electricity during ebbs in solar and wind production. It trails Eavor and Fervo by several years, and has so far not built a demonstration model. 

“Hotter, deeper, faster, cheaper” 

So far, none of these companies have yet to drill into rock hot enough to maximize the technology’s full potential. Fervo claims to have hit temperatures of 375°F, while a representative from Eavor said that the company’s experimental deep drilling operation in New Mexico has reached rock as hot as 480°F. Both are well below the potentially industry-changing 700°F point at which energy capacity and efficiency would increase dramatically, enabling the technology to create enormous quantities of power. 

Being able to access rock at ever higher temperatures is a fundamental feature of next-generation geothermal because the amount of energy contained in water increases non-linearly as temperatures rise: If you increase the water temperature from 200°C (390°F) to 400°C (750°F), explained Avaque, the amount of power you can generate at the surface theoretically increases by a factor of 10.

Getting there, of course, is difficult, in part because of the need to develop the tools necessary for this kind of drilling. In this regard, next generation geothermal has received significant financial and technical assistance from the oil and gas industry, accelerating progress. Fervo is backed by Devon Energy, Sage by Chesapeake Energy and Eavor counts BP and Chevron as backers. While drilling beyond temperatures of 200°C is not typically necessary to extract oil and gas, the industry has been collaborating with geothermal projects to facilitate drilling into hotter rock. Utah FORGE has collaborated with the industry to strip out the most heat-sensitive components of drilling rigs, for example, and Eavor has worked extensively with the industry to develop new drilling techniques. In particular, it has created a new insulated drill pipe system that can carry coolants miles down wells so that drill heads and electronics stay cool enough to function. These advances, said Evaor CEO John Redfern, can enable otherwise conventional equipment to drill into rock at temperatures in excess of 570°F.

But there are significant barriers to drilling into rock that is 700°F or more because pumping coolants down the well — especially once a well extends many miles deep — no longer works. “That trick runs out of power as you go deeper,” said Araque, “because the journey is longer,” meaning that too much heat conducts into the coolant. 

So companies are developing advanced drilling tools that don’t rely on conventional rotary methods. Quaise, the startup led by Araque, is using high-energy microwaves to vaporize rock with the hope of drilling holes beyond 10 miles in depth. “It’s brute force,” said Araque. GA Drilling from Slovakia, meanwhile, is building plasma torches to do something similar. And Redfern, without elaborating, said that Eavor is developing its own technology to achieve the same goal. The idea, said Araque, is that conventional drilling would take a well down to a depth of one to three miles before the new style of drill head is lowered into the hole to tackle the hot rock. Playing a supporting role, said Boyd, are robust new electronics systems, inspired by those used for space exploration, to enable monitoring systems to continue working at high temperatures.

The approaches are all designed to reach the hottest rock possible while keeping total drilling times and costs as low as possible. “Hotter, deeper, faster, cheaper,” said Redfern. “That’s our motto.”

High rewards, high risks

The biggest challenge facing next-generation geothermal may not be technical, but financial. The prototype facilities being developed by the likes of Eavor, Fervo and Sage are fantastically expensive: A representative for Eavor said the company has raised over $680 million in equity investment debt and grants, and according to Redfern is spending $15 million a month. “The execution is expensive, because you're playing with the subsurface at great depths,” said Araque. “You'll do nothing with $1 million. You do a little bit with $10 million. You're in the sweet spot at $100 million to $1 billion.” 

The cost of drilling accounts for about “50 percent of your total development costs,” said Joseph Moore, a principal investigator at Utah FORGE. All the experts Aventine spoke to agreed that driving down the cost of drilling the wells will help make the technology viable. Those costs are highly correlated with how long it takes to drill the hole: “It's all the time-related issues that are going to help us reduce those costs,” said Boyd, pointing out that many other costs are relatively fixed. Progress is being made on cutting drilling time: The demonstrator at Utah FORGE showed an increase in drilling speed of over 500 percent over the course of three years. But the DOE still predicts that the industry needs $20-$25 billion of investment to prove the market opportunity, and as much as $250 billion to roll it out at scale across the U.S. Yet despite the success of some startups — including Eavor and Fervo, which has also raised a little over $400 million — raising sufficient funds is difficult. “Capital is very hard to get at relatively high levels of risk,” said Araque.

Ultimately, it may be the success or failure of the early demonstrators that cements the future of the technology. Redfern admitted that there is huge pressure on early plants to perform. They don’t have to be perfect, but equally they cannot fail. “We have to make sure that we're making fast enough progress,” he said. “If you trip, you’re in trouble.”

Artist rendering of Berge Olympus, ( owned by Berge Bulk ) a cargo ship equipped with four sets of WindWings, built to set sail in 2025. Berge

High-tech sailboats could cut marine emissions. Harnessing the wind as a means of propulsion on the seas may enjoy a renaissance, due to a push to decarbonize freight and the appearance of a variety of modern sails that increase the efficiency of existing boats. The Economist reports that several designs currently in use — none of which look much like traditional sails — are taking advantage of well-established aerodynamic principles to harness energy from wind. The designs are varied: Flettner rotors are essentially large vertical cylinders whose rotations interact with winds to create pressure differentials that give rise to thrust; suction sails pull in air through large vertical tube-like structures to generate propulsion; and so-called WindWings are more akin to vertical airplane wings, which catch the wind to generate motion the way an aircraft generates lift. These models can be retrofitted to existing boats like ferries and container vessels to reduce the need for fuel by 5 to 25 percent. But they are not cheap: Norsepower, a Finnish company, is selling its Flettner rotors for more than $1 million apiece, a sum that the company says will be recovered in fuel savings over the course of three to 10 years. Yet despite the cost, the technologies could prove vital, since the maritime industry is committed to reducing emissions and there are currently few other practicable solutions for large, long-haul vessels available at commercial scale.

Apple’s big, bold bet on AI. Apple has been criticized as a laggard when it comes to artificial intelligence. Given its frequent marketing claims of strong user security and privacy — and the criticism that the current slate of generative AI systems has received for data intrusions, hallucinations and more — it’s perhaps not surprising that Apple has hung back. But that changed in June when the company announced Apple Intelligence, an offering that it hopes will allow it to deploy cutting-edge AI systems to users while maintaining privacy and quality. To accomplish this, Wired reports, Apple has erected a number of guardrails. For example, the company will use models that reside on the user’s hardware, meaning that personal data doesn’t have to be transferred to the cloud for routine tasks. For more difficult tasks, Apple will use a system called Private Cloud Compute that it claims will use cloud server hardware built using secure Apple processors. Apple also claims to have carefully curated the data it uses to train its generative models to reduce the frequency of hallucinations, and that its collaboration with OpenAI, which will allow Siri and Apple’s writing tools to use ChatGPT, will require users to confirm they’re happy using the third-party service. Still, a big question hangs over Apple's AI deployment: Is this enough? The rise of LLMs has taught us that no amount of testing inside a company can uncover everything that can go wrong with a generative AI system; widespread deployment does a much better job. Apple’s big bet is that it's done enough to maintain its reputation for privacy and quality as it strives to keep up with the pack. 

An AI-powered black box for the OR. Flight recorders are widely acknowledged to have improved aeronautical safety by providing greater after-the-fact understanding of how accidents happened. A startup called Surgical Safety Technologies thinks it can replicate that success inside hospital operating rooms. As MIT Technology Review reports, the company uses audio captured by microphones, video from dome cameras and data taken from medical devices inside operating rooms and feeds it all into a number of AI algorithms for processing. One of the results is a summary of each medical procedure, including highlight videos and statistics about what happened in the room — from the surgical instruments used to the adherence to surgical safety checklists, which are designed to minimize complications. The technology — which costs tens of thousands of dollars to install and a similar amount to employ annually — is already being used in almost 40 institutions across the U.S., Canada and Europe. Perhaps the biggest obstacle to its wider adoption is the reaction of surgeons and nurses who would be recorded by it; many have simply refused to work in operating rooms outfitted with the technology. Surgical Safety Technologies has tried to address concerns about employee privacy by obscuring faces and distorting voices in the videos it obtains, but hospital administrators can still determine when and where an operation took place, making identification pretty straightforward. Ultimately, the technology’s impact on safety could be the determining factor, and the company claims that it will soon publish research showing that the presence of the devices in ORs has led to improvements in outcomes, including reduced ICU stays and mortality.

Syringe inserted into a vial with HIV vaccination. Shutterstock

For nearly 40 years, researchers have been trying and failing to create an HIV vaccine, a sort of holy grail for many researchers because the virus has so many unusual characteristics. Now, after decades of failure, the puzzle has started to look a lot more solvable thanks to an approach that most researchers believe is likely more promising than any tried in the past: Creating what are known as “broadly neutralizing,” antibodies, which can successfully act against the HIV virus in its many mutations instead of just one. (HIV mutates so often within one person that traditional vaccines aimed at a single mutation, like those for Covid or influenza, are not effective.) Building up those broadly neutralizing antibodies in a person is a complicated and time-consuming process and researchers are still not certain how it would work or what a vaccine regimen for HIV would look like. However, a handful of possible approaches currently in animal trials and very early human trials look promising enough to shed some light on those questions. 

But just as this transformative vaccine can be glimpsed on the horizon (albeit at least 10 years away, say experts), its importance as a goal is being challenged by the overwhelming success of a parallel strategy for preventing HIV infection: pre-exposure prophylaxis, or PrEP drugs. While the drugs, which have been available as a daily pill to prevent HIV disease for years, are highly but not 100 percent effective, a new injectable version now seems to meet the 100 percent threshold. In late June Gilead Sciences announced the results of a large clinical trial in which an injectable version of PrEP — administered twice a year — prevented all 2,134 participants from contracting the virus. 

While not technically a vaccine, in many ways this drug meets a vaccine’s primary goal: preventing infection without significant effort on the part of the person taking the drug. So the Gilead news raises fresh questions about what a successful vaccine would look like and what it might take to get there. 

While rates of HIV infection have declined due to education and preventative measures, the virus still affects millions of people globally. About 39 million people were living with HIV in 2022, and more than one million were infected in that year alone. While consistent data on PrEP uptake varies globally, in the United States only about one-third of people who could benefit from PrEP are prescribed it, which is not enough for the U.S. to reach its goal of reducing new HIV infections by 90 percent by 2030. 

Aventine spoke with five researchers about what an HIV vaccine would mean and how to balance its benefits against those of other already available therapies. While most experts agreed that preventative drugs like PrEP shouldn’t be compared to a vaccine because access to each kind of drug varies so widely, there was an acknowledgment that the new twice-a-year treatment from Gilead changes the stakes for a vaccine. What follows are their comments in full. 

Artist rendering of the James Webb Telescope in orbit with its sunshields in place. NASA

1. Somewhere between Earth and Mars. As experiments go, NASA’s James Webb Space Telescope, launched in December 2021 to view the farthest reaches of the universe, is an expensive one, costing U.S. taxpayers $10 billion. That is why Raytheon, the defense contractor that built the JWST’s software and also controls its flight operations, isn’t taking any chances when it comes to monitoring its deployment. MIT Technology Review explains that this is made possible by technology known as a digital twin — a computer-based model of the JWST that is updated with 800 million data points every day to create a video of the telescope in just about real time. The simulation can also be used to predict how the real telescope will respond to new instructions and software updates: Engineers simply use a copy of the digital twin to test out new programming. The insights gained from the JWST’s digital twin are now being used in some of Raytheon’s other projects — such as the development of radar systems, missiles and airplanes — to create simulations so that those projects too can be more closely monitored. 

2. Hellisheiði, Iceland. On an expansive open plain a short drive from Reykjavík sits the world’s largest ​direct air capture plant, called Mammoth and built by the startup Climeworks. Designed to suck CO2 out of the air so that it can be safely sequestered, the facility, which went online last month, uses huge racks of electric fans to suck in air and run it over solid filters that trap and collect carbon dioxide. It then passes that CO2 on to a partner called Carbfix, which deposits it inside basaltic rock underground. The whole process is powered by geothermal energy. The company claims that Mammoth, which is 10 times larger than an earlier direct air capture plant Climework built, is capable of removing 36,000 tons of CO2 from the air each year at half the cost of the smaller plant. This is notable progress toward making direct air capture (or DAC) more affordable, but there’s still a significant way to go: The startup spent “low-triple-digit millions” on building the plant, according to Canary Media, and the cost of removing a ton of CO2 is nowhere close to the $100 that is considered to be a commercially viable price point. The company argues that its DAC facilities serve the larger goal of building even bigger facilities in the future, including one in Louisiana that could remove over a million tons of carbon annually. To put that number in perspective, the International Energy Agency estimates that DAC facilities will need to remove around 65,000,000 tons of CO2 per year by 2030 in order to reach net zero goals by 2050. 

3. Santiago, Chile. For more than a decade, Chile has been struggling with drought that has caused megafires and dried out the wells of subsistence farmers. The main culprits are huge eucalyptus and Monterey pine plantations that suck up water, along with avocado farming for export markets. But as this story from Rest of World describes, locals are increasingly concerned about the impact of data centers on water scarcity. There are currently 22 data centers in Chile, and an announcement earlier this year by the nation’s president, Gabriel Boric, suggested that the government hopes to add another 28. But their impact, according to local residents and environmental researchers, can be devastating to the water supply. One data center proposed by Google and approved in 2020 planned to extract 228 liters of water per second from underground wells — or more than 7 billion liters each year. That project is now on hold because of protests that resulted in an investigation by Santiago’s environmental tribunal, which determined that Google must reassess its environmental impact before proceeding. According to Rest of World, campaigners are buoyed by their success against Google, but many other projects — including Amazon’s first data center in the country — look set to continue for now.

The messy quest to replace drugs with electricity, from MIT Technology Review
5,700 words, or about 21 minutes

Did you know that when your skin is injured it produces an electric field that’s strongly believed to encourage healing? In fact, the human body both induces and responds to electric fields of all kinds, which is in part what created hope about a decade ago for a new breed of treatments called electroceuticals that would theoretically help replace conventional drugs. But, as this feature from MIT Technology Review explains, progress has been slow despite the fact that somewhere in the region of a billion dollars has been invested in the idea. While hopes for electricity-based treatment aren’t dead, new approaches are far more complex and nuanced than earlier ones.

The AI revolution is coming to robots. How will it change them? from Nature
2,600 words, or about 10 minutes

The surging success of generative AI has a lot of people thinking hard about how the techniques might be carried over to the world of robotics, which is having its own — albeit less hyped — moment of accelerated technological advance. Yet while there’s scope for roboticists to take advantage of new forms of AI, this story makes one thing very clear: There just isn’t enough high-quality data on which to train AI for robots. There are attempts to change this situation, including collaborations to pool data captured by labs around the world, attempts to create synthetic data using simulations and even simply having robots learn by watching videos of humans performing tasks. But the conclusion of this story is clear: None of these solutions is quite sophisticated enough for robotics to have its own ChatGPT moment just yet.

The Age of the Drone Police Is Here, from Wired
5,600 words, or about 20 minutes

Way back in 2018, Chula Vista, California, became the first city in the U.S. to start a so-called Drone as First Responder (DFR) program, which would dispatch quadcopters with cameras to some of its 911 calls. Since then, the practice has become an integral part of the city’s policing, according to this investigation by Wired. Between July 2021 and September 2023, drones were dispatched to 7 percent of the city’s 911 incidents, including about half of those that involved people with weapons. In total, Chula Vista’s police drones have completed around 20,000 flights, and the police department claims they have helped shorten response time and even save lives. But perhaps the most interesting element of this story is the tension experienced by the city’s residents: In a survey, the majority said that they worried about how the devices may be used to record people who weren’t involved with a crime; at the same time a majority were in favor of the drones being used by police.